Transient Performance Enhancement Of Adaptive Control Systems And Its Application To Aircraft Attitude Control

Aircraft systems,as typical multi-input multi-output and nonlinear mechatronics systems with strong coupling,usually suffer from uncertainties,such as high-frequency unmodeled dynamics and unknown external disturbances,and these factors will increase the complexity of attitude control system design.To address the parameter uncertainties in these systems,adaptive control provides an effective solution,whose advantage is that the model parameters or controller parameters can be updated online according to the control system error to adapt the system variations or external disturbances.However,although some of theoretical adaptive control methods have been applied to the attitude control of different aircraft systems,the slow control response,high-frequency oscillations,and sluggosh transient period in the adaptive attitude control still restrict the practical application of adaptive control to electromechanical systems such as aircrafts.In viewing the existing adaptive control literatures,it is found that the most of adaptive control designs mainly focus on the steady-state performance of closed-loop systems and the convergence of system tracking errors,while quantitative analysis on the transient performance of closed-loop systems(such as rise time,settling time,overshoot,etc.)is still a difficult yet open problem to solve in the control community.From the perspective of practical application,good transient performance(e.g.,short rising time and small overshoot)is an important prerequisite for the system to enter the steady-state safely and reliability.Hence,aiming to address the requirements of aircraft attitude control,we will explore several effective ways to enhance the transient performance of adaptive control systems with uncertainties,where the model reference adaptive control(MRAC)framework is considered.The thesis proposes three methodologies in corresponding to the three major elements of adaptive control systems(i.e.,adaptive law,reference model and controller structure)to improve the transient performance.Finally,these derived theoretical results are used to address the aircraft attitude control problem.To carry out experiments,a typical 3-degrees of freedom(DOF)aircraft prototype is designed and constructed in our lab.Finally,extensive simulations and experiments are carried out to verify the effectiveness of these proposed methods.The main research work of this thesis are summarized as follows:(1)The design and integration of a 3-DOF prototype aircraft system.The considered aircraft system is a simple prototype,which can be used to simulate the motion of aircrafts with three main degrees of freedom(i.e.,pitch,roll and yaw).We first introduce the design and integration of this semi-physical platform,and establish its mathematical model.Then,this derived model is used as a representative to analyze the factors affecting the transient performance of adaptive control systems and clarify the conservativeness of the existing well-known methods.Finally,Cauchy-Schwarz inequality and L₂-norm are introduced to analyze the transient performance of adaptive control systems,and several effective ways that can be used to enhance the transient performance are suggested.(2)To address the slow convergence or non-convergence of parameter estimation in the classical adaptive control methods,a novel adaptive law containing the parameter estimation error is proposed.First,the explicit formulation of parameter estimation error is derived by using filter operations and introdcing auxiliary variables,and then they are used to construct a new leakage term,which is imposed into the gradient descent adaptive laws.This composite adaptive law deriven by both the parameter estimation error and system tracking error can ensure rapid convergence of estimate of system model parameters,so as to improve the transient performance of control systems.Finally,this method is further extended to adaptive control for systems with unknown control input matrix,where a projection algorithm is derived.(3)To tackle the sluggish response and large overshoot of adaptive control caused by the ineffective compensation of uncertain dynamics in the adaptive control systems,a closed-loop model reference adaptive control with a new external command governor is proposed.The performance of adaptive control system is analyzed based on the Cauchy-Schwarz inequality.To speed up the convergence of the system tracking error and suppress the peak phenomenon induced by the classical closed-loop reference model,an auxiliary term based on the tracking error is constructed and then added to the external command signal to construct a new external command governor to compensate the effects of uncertain dynamics in the MRAC systems.Finally,this derived external command governor is extended to the adaptive control systems with unknown control input matrix.(4)To ahndle the performance degradation of adaptive control induced by the high-gain learning or high-frequency contents,a new adaptive control with a constructive frequency selection mechanism is proposed.First,an extraction strategy of high-frequency dynamics via a low-pass filter is introduced to obtain the high-frequency information embedded in the closed-loop system.Then the derived high-frequency contents are introduced into the controller through a feedback mechanism,such that the high-frequency contents induced in the closed-loop control system can be effectively compensated.With the help of this compensator,a fairly large leaning gain can be used in adaptive control system to achieve fast adaptation without triggering high-frequency oscillations.Apart from the above theoretical research,this thesis also considers two types of typical aircraft systems(i.e.,3-DOF aircraft prototype and a wing rock dynamics model)to carry out simulation.Experimental verification based on this constructed 3-DOF semi-physical platform are also carried out.Both simulation and experimental results show that the control schemes that are derived in corresponding to the learning mechanism,reference model and control structure can improve the performance for aircraft attitude control.In this respect,the thesis provides potential pathways to stimulate the application of adaptive control to actual electromechanical systems.